The present invention relates to a method of, and an apparatus for, reallocating logical disk devices to physical disk devices using a storage controller, and in particular, to a method of, and an apparatus for, reallocating logical disk devices to physical disk devices using a storage controller for improving data access performance.
A technology related to data arrangement on a disk array has been disclosed in pages 109 to 116 of an article “A Case For Redundant Arrays Of Inexpensive Disks (RAID)” reported by D. Patterson, G. Gibson, and Randy H. Katz in ACM SIGMOD conference held at the University of Illinois, Chicago in June 1988. In this article, the disk arrays are classified into five levels including RAID1 to RAID5.
In a disk array of RAID1, write data from a data processor unit is written in a main disk device and a copy thereof is stored in a sub-disk device called a mirror, thereby guaranteeing reliability of data in RAID1. In RAID1, since check information is a copy of the original data, only a small overhead is required to generate the check information and hence a desirable access performance can be obtained. However, the usage efficiency of physical storage devices is reduced to, for example, 50%.
Additionally, in the disk array of RAID5, check information called parity is produced for a plurality of write data items from the data processor unit. In RAID5, in a parity creation to generate a new parity item, it is necessary to read data and parity respectively generated before an update operation. Therefore, lengthy overhead is required for the generation of check information and hence the access performance is reduced. However, only one parity item is created for a plurality of data items, the usage efficiency of storage devices is higher when compared with that of storage devices of RAID1.
Furthermore, in JP-A-7-84732 (U.S. Priority 109,137 (Aug. 19, 1993), there has been disclosed a method in which disk devices are subdivided into RAID1 and RAID5 areas such that data is moved between RAID1 and RAID5 according to a performance protocol.
Moreover, in a report “Performance Evaluation Of Hot Mirrored Disk Arrays On Disk Failure” written by K. Mogi and M. Kitsuregawa in pages 19 to 24 of the Technical Report of IEICE, CPSY95-82, DE95-68 (1995-12), Vol. 95-No. 407, there has been disclosed a technology in which a RAID level of data is dynamically changed according to an access frequency. Specifically, disk devices are subdivided into RAID1 and RAID5 areas such that data for which write access is requested is preferentially stored in the RAID1 area. With this provision, data having a high access frequency can be stored in the RAID1 area and data having a low access frequency can be stored in the RAID5 area.
As a result of, this technology, physical disk devices having mutually different values of storage capacity and physical disk devices having mutually different RAID levels can be installed in a mixed fashion in the storage subsystem. Furthermore, data in a logical disk device can be stored in arbitrary logical disk devices according to indices such as the access frequency and pattern thereof. In addition, when storing data having a high access frequency, it may also be possible to alter the storage position thereof so that the data is written in a physical disk device having a higher speed.
According to the above, the data storage position is varied in the unit of data to be accessed. Therefore, successive data items in a logical disk device which is directly accessed by the data processing unit are discontinuous in physical disk devices in which the data items are actually stored.
On the other hand, according to the report “DE95-68” above, each time a write request is issued, data regarded to have a low access frequency is moved from the RAID1 area the RAID5 area so as to write the data in an available area thus reserved in the RAID1. Consequently, when the access frequency is low for an access pattern in a random access operation, most of data items moved to the RAID1 area are returned again to the RAIDS area.